Devlog #22

The long DEVSIM run finally finished. It took about 6 hours and matched the expert result almost exactly, which closes out the scoreboard phase.

Then I started the next phase, which is about AC. That essentially means wiggling the voltage back and forth and watching how the device reacts at different speeds. Before, my simulator could only handle super slow wiggles, or fast ones where I froze some electrons by hand. Now it can handle any speed.

To get there, I first taught the simulator how fast electrons and holes pile up or drain away over time. Then I wrote the part that works out how much current flows into each contact when one of them gets wiggled.

I tested it on diodes, since they have textbook formulas to compare against. A long diode matched within about 1%, and a short one within 1.3%. Along the way I found a bug where my mesh maker gave up on very long devices, so I fixed that too.

The MOS capacitor is giving me some trouble. At really slow wiggles, the math needs more precision than normal computer numbers have, so the answers came out wrong by up to 46%. Fast wiggles work fine. Right now I’m trying a trick where I make the electrons react faster, so the slow case happens at a speed the computer can handle.

Below are some of the DEVSIM scoreboard values